VCPI Releases DRUG-seq Profiles for the ReFRAME Library: 13,000 Clinical-Stage Compounds, Open Under CC BY 4.0
The Virtual Cell Pharmacology Initiative's largest release to date: arrayed DRUG-seq transcriptomics for the ReFRAME drug repurposing collection — 12,949 compounds, screened at 10uM dose, one compound per well, in THP-1 cells. Free to every registered user at thevirtualcell.com.
When we launched VCPI, the promise was simple: pharmacology-grade transcriptomic data, generated on a standardized platform, released to the community for free. Today's release is the biggest step toward that promise so far. We have profiled the full ReFRAME library — 12,949 compounds — by DRUG-seq, and the data is now live.
Why ReFRAME
ReFRAME (Repurposing, Focused Rescue, and Accelerated Medchem) was assembled by Calibr-Skaggs at Scripps Research, with support from the Bill Gates Foundation and The Wellcome Trust, as the most comprehensive collection of drug-like molecules that have already reached clinical development or undergone significant preclinical profiling. Roughly 13,000 compounds, drawn from commercial drug databases and patent filings, many synthesized from scratch because they were not commercially available.
Our DRUG-seq data on the ReFRAME library is a unique resource for the community. Virtual cell models are meant to predict how human cells respond to drugs. To learn that, they need measurements on molecules with well-characterized mechanisms, known targets, human-relevant chemistry, and the kind of structural diversity that a general-purpose model has to handle. ReFRAME delivers all of that in a single, well-annotated library. It also comes with over a decade of published phenotypic screening results at reframedb.org, which means every transcriptomic profile we release can be cross-referenced against existing biology.
What's in the release
- 12,949 compounds profiled by arrayed DRUG-seq. Each compound is treated in its own barcoded well. One compound, one well, one clean transcriptomic readout
- Top dose, with dose response follow-up coming soon. Screened at one dose per compound [10 µM], two replicates, 24-hour treatment, with transcriptional and cytotoxicity controls on every plate. We will publish data on a six point dose response on the top 10% of hits soon.
- THP-1 cells, the VCPI reference backbone. The same human monocytic line is used across every VCPI release, so this dataset is directly comparable to the 14,000+ compounds already released on the platform.
- ~10,000 genes recovered per well at 2M reads/well.
- Open under CC BY 4.0. Download through your browser or pull via the VCPI Python client. Commercial use included.
With this release, VCPI now holds DRUG-seq profiles for more than 26,000 compounds in a single cell line under one standardized protocol — the largest public arrayed chemical-perturbation transcriptomics resource available.
A multilayer perceptron mechanism of action classifier trained on the ReFrame compounds (colored dots) allows prediction of mechanisms of action of unannotated compounds from previous VCPI releases (grey dots).
Why this matters for virtual cells
The field has no shortage of perturbation data. What it lacks is perturbation data that reflects pharmacology: dose, replicates, drug-like chemistry, and a consistent biological background. ReFRAME with DRUG-seq fills a specific gap:
- Mechanism coverage. Thousands of compounds with annotated targets and clinical history give models a labeled scaffold to learn structure-to-response relationships, not just correlations between unknowns.
- A real benchmark. Because ReFRAME is publicly annotated and widely screened, it is an obvious held-out set for evaluating whether a model generalizes to new chemistry — the test the field keeps asking for.
- Repurposing, directly. A full-library transcriptomic signature of clinical-stage molecules is a repurposing resource in its own right. Connectivity-style analyses against disease signatures are now possible at drug-like scale, on data that was generated in one place, under one protocol.
Get the data
Create a free account at thevirtualcell.com, log in, and download. The ReFRAME release sits alongside all previous VCPI datasets. If you have compounds of your own, submit them and we will run them through the same DRUG-seq pipeline at no charge, with an embargo option for IP-sensitive molecules.
Get the ReFRAME dataset at thevirtualcell.com.
Thanks to the Calibr-Skaggs Institute for Innovative Medicines at Scripps Research for making the ReFRAME library available to the initiative, and to the VCPI Superusers whose compound submissions and feedback shaped this release.